Research On Monitoring Red Tides In Qinhuangdao Sea Area Using Remote Sensing Images

Red tides refers to an ecological abnormal phenomenon that some phytoplankton, protozoa, or bacteria in seawater under certain environmental conditions multiply explosively or get together causing discoloration of water. This article tries to complete the red tides monitoring in Qinhuangdao Sea from two aspects. Firstly, this article bulids neural network model based on field measured data to predict the red tides; Secondly, using MODIS satellite remote sensing data, this article builds model aim at the detection of red tides in Qinhuangdao Sea and monitors its development trend.In aspect of building red tides forecast model based on neural network with field measured data, this article applies genetic algorithm (GA) and particle swarm optimization algorithm (PSO), sets up PSO-LMBP and GA-LMBP model to forecast the red tides, to avoid falling into local minimum, which LMBP network may easily occur in the training. The simulation experiments show that the improved algorithm effectively improves the stability and accuracy of the red tides forecast. In addition, this article also builts GRNN neural network model based on principal component analysis algorithm (PCA) to predict the red tides. Experiment results show that the PCA-GRNN red tides forecast model has high prediction accuracy and fast network convergence.In aspect of monitoring red tides using MODIS remote sensing data, this article respectively uses chlorophyll a concentration threshold segmentation method, band ratio threshold segmentation method, maximum likelihood supervised classification method and decision tree method to extract red tides information of Qinhuangdao Sea by ENVI software. Experiment results show that the red tides monitoring effect in Qinhuangdao Sea area is general through method of chlorophyll a concentration detection model, this model remains to be improved; the band ratio threshold segmentation model and maximum likelihood supervised classification have some shortcomings; decision tree red tides detection model has achieved satisfactory results when detecting red tides in Qinhuangdao Sea area.